Psychiatric patient behavior evaluation system
By introducing Gaussian mixed model and adaptive joint clustering in the behavioral assessment system of psychiatric patients, combining dynamic learning rate and adaptive change order factors, the system's poor adaptability to complex behavioral data and low parameter correction efficiency is solved, and the system's ability to adapt to complex behavioral data is achieved. More accurate and fast behavioral assessment.
Patent Information
- Application Number
- CN202411397295.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-09
- Publication Date
- 2025-01-10
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing psychiatric behavioral assessment system cannot capture the potential complex relationships in the behavioral data of psychiatric patients, resulting in poor adaptability and low evaluation accuracy, and inappropriate initial parameter setting leads to low parameter correction efficiency, affecting the evaluation effect.
Introduce hidden variables for row clustering and hierarchical clustering based on Gaussian hybrid model, combine adaptive joint clustering, update parameters using dynamic learning rate, and optimize system parameters through adaptive variable order factors and higher-order update strategies to avoid local optimality .
It improves the system's robustness and evaluation accuracy of behavioral data, enhances parameter optimization accuracy and stability, and ensures the rapid and accurate behavioral evaluation of mentally ill patients.
Smart Images
Figure CN119296732A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a behavior assessment system for psychiatric patients. Background Art
[0002] The psychiatric patient behavior assessment system helps psychiatrists better understand patients' emotions, cognition and behavior patterns through comprehensive analysis of patients' daily behaviors, psychological states and environmental factors. However, the general psychiatric patient behavior assessment system is unable to capture the potential complex relationships in the behavioral data of psychiatric patients, and is limited to observing surface data, which makes the system poorly adaptable to the complex behavioral data of patients, and thus leads to low accuracy in behavioral assessment. The general psychiatric patient behavior assessment system has problems such as improper initial parameter settings, low search efficiency during parameter adjustment, and weak global exploration capabilities, which leads to low efficiency in the system parameter correction process and ultimately poor results in patient behavior assessment. Summary of the invention
[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a psychiatric patient behavior assessment system. In view of the fact that the general psychiatric patient behavior assessment system cannot capture the potential complex relationships in the behavior data of psychiatric patients and is limited to observing surface data, which makes the system poorly adaptable to the complex behavior data of patients, and thus leads to low accuracy of behavior assessment, this solution introduces hidden variables to perform row clustering on the behavior data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and utilizes the association between patient behavior characteristics and behavior risk levels based on adaptive joint clustering to improve the system's robustness to behavior data and the accuracy of behavior assessment results, and introduces dynamic learning rate update parameters to improve The system has convergence when processing the behavioral data of psychiatric patients, which enables the system to evaluate the patient's behavior more quickly; for the general psychiatric patient behavior assessment system, there are problems such as improper initial system parameter settings, low search efficiency during parameter adjustment, and weak global exploration capabilities, which lead to low efficiency of the system parameter correction process and ultimately poor effect on patient behavior assessment. This solution introduces an adaptive variable order factor, dynamically adjusts based on an iterative process, enhances search efficiency, introduces a high-order update strategy, improves the accuracy and stability of system parameter optimization, avoids falling into local optimality through perturbation factors, and thereby improves the system's processing speed for psychiatric patient behavior data, ensuring that the final system's accuracy in patient behavior assessment is high enough.
[0004] The technical solution adopted by the present invention is as follows: A psychiatric patient behavior assessment system provided by the present invention includes a data acquisition module, a behavior data matrix construction module, an adaptive joint clustering module and a psychiatric patient behavior assessment module;
[0005] The data collection module collects historical mental patient behavior data and real-time mental patient behavior data;
[0006] The behavior data matrix construction module constructs a behavior data matrix based on the collected data;
[0007] The adaptive joint clustering module introduces hidden variables to perform row clustering on the behavior data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and performs adaptive joint clustering. After clustering correction, the parameters are updated by introducing a dynamic learning rate; an adaptive variable order factor is introduced, dynamically adjusted based on an iterative process, and a high-order update strategy is introduced to optimize the initial parameters of the model, thereby realizing clustering of the behavior data;
[0008] The psychiatric patient behavior assessment module implements the assessment of real-time psychiatric patient behavior data based on clustering results.
[0009] Furthermore, in the data collection module, the historical psychiatric patient behavior data and the real-time psychiatric patient behavior data both include emotional state, sleep duration, social interaction, physiological indicators, physical activity level and anxiety level; the historical psychiatric patient behavior data also include behavioral risk level; the behavioral risk level is used as a data label, and the label is only used to define the clustering label and does not participate in the clustering process.
[0010] Furthermore, the behavior data matrix construction module converts the collected data into vectors and performs standardization processing, and then constructs a matrix, in which the rows of the matrix represent patients and the columns of the matrix represent the behavioral characteristics of the patients; obtains the behavior data matrix; and defines the clustering label as the label with the largest number of historical data in the cluster to which it belongs.
[0011] Furthermore, the adaptive joint clustering module specifically includes the following contents:
[0012] Row clustering unit: Row clustering is to cluster the rows of the data matrix, cluster the patient rows into H behavioral risk level groups based on the Gaussian mixture model, and identify the differences in behavioral risk levels between different patient groups. The probability of row clustering is expressed as: ; The distribution of hidden variables is expressed as: ; The distribution of the behavioral data of mental patients in the context of row clustering is expressed as: ; where p(·) is the probability; p(·|·) is the conditional probability; vhr is the probability that patient r belongs to behavioral risk level h; τ h is the population proportion of the hth cluster, indicating the proportion of patients belonging to this cluster; whr is a hidden variable vector, indicating the hidden variables associated with patient r; N(·) is a Gaussian distribution, is the identity matrix; mr is the behavioral data vector of patient r; is the mean vector of the h-th cluster, representing the feature center of the cluster; is the weight loading matrix, describing the relationship between behavioral features and hidden variables; is the diagonal covariance matrix of the hth cluster, representing the variance of independent errors; H is the total number of behavioral risk level groups;
[0013] Column clustering unit; row clustering is to cluster the columns of the data matrix, cluster the behavioral features into L feature categories based on hierarchical clustering to identify feature groups with similar behavioral risk levels, and define the behavioral feature clustering as ; Calculate the covariance matrix after column clustering, expressed as: ;in, is an indicator matrix, which is 1 if the rth row belongs to the lth column cluster, and 0 otherwise; is the covariance matrix after column clustering; is a diagonal matrix containing the characteristic covariance of column clustering; T is the matrix transpose;
[0014] Adaptive joint clustering unit; combining row clustering and column clustering, the probability density of the patient's behavioral data is finally expressed as: ;
[0015] Cluster correction unit: updates the posterior probability of patient row clustering, expressed as: ; Update the posterior probability of the hidden variable, expressed as: ; The model parameters are optimized by maximizing the logarithmic expectation, expressed as: ; Where E[·] is the expected value of the behavioral risk level; Maximize is the maximization operation; p(·; ·) is the joint distribution probability; are model parameters; is the initial parameter;
[0016] Parameter updating unit; update population ratio, expressed as: ; Update the mean vector, expressed as: ; Update the covariance matrix, expressed as: ;in, and are the population proportions of the h-th cluster after and before the update, respectively; and are the mean vectors after and before the update respectively; and are the covariance matrices before and after the update; α is the dynamic learning rate, , α 0 is the initial learning rate, T1 is the maximum number of iterations, is a smoothing term; R is the total number of patients;
[0017] Clustering determination unit; the maximum number of iterations and clustering threshold are pre-set. When the average silhouette coefficient of all samples is higher than the clustering threshold when clustering converges, clustering is completed; otherwise, if the maximum number of iterations is reached, the initial parameters are corrected; otherwise, the adaptive joint clustering unit is returned to continue iteration;
[0018] Initial parameter correction unit; specifically including:
[0019] Construct parameter space; construct parameter search space based on initial hidden variable vector, mean vector, covariance matrix, and weight loading matrix; randomly initialize the search individual position, and use the average silhouette coefficient of all samples when clustering convergence is reached based on the individual position as the individual fitness value;
[0020] Position update; introduce adaptive variable order factor; the formula used is as follows:
[0021] ;
[0022] ;
[0023] ;
[0024] Where v is the adaptive variable order factor at the tth iteration; t is the current iteration number, and tmax is the maximum iteration number; is the position of the jth dimension of the uth individual; f u is the individual fitness value; is the position of the jth dimension of the global optimal individual; γ is the attenuation coefficient; γ0 is the initial attenuation coefficient; is the disturbance factor; randn is a standard normal distribution random number; is the decay rate;
[0025] Search judgment; when there is an individual fitness value higher than the clustering threshold, the initial parameter correction is completed, the individual position is the optimal initial parameter, and the clustering result is obtained; if the maximum number of iterations is reached, return to the construction parameter space: otherwise update the fitness value and return the position update.
[0026] Furthermore, the psychiatric patient behavior assessment module is based on clustering results, and uses the cluster label to which the real-time psychiatric patient behavior data belongs as the assessment result of the real-time psychiatric patient behavior data.
[0027] The beneficial effects achieved by the present invention using the above scheme are as follows:
[0028] (1) The general psychiatric patient behavior assessment system is unable to capture the potential complex relationships in the behavioral data of psychiatric patients and is limited to observing surface data, which makes the system poorly adaptable to the complex behavioral data of patients, and thus leads to low accuracy of behavioral assessment. This solution introduces hidden variables to perform row clustering on behavioral data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and uses the association between patient behavioral characteristics and behavioral risk levels based on adaptive joint clustering to improve the system's robustness to behavioral data and the accuracy of behavioral assessment results. By introducing dynamic learning rate update parameters, the system's convergence in processing psychiatric patient behavior data is improved, thereby enabling the system to evaluate patient behavior more quickly.
[0029] (2) In view of the problems of improper initial parameter settings, low search efficiency during parameter adjustment, and weak global exploration capabilities in general psychiatric patient behavior assessment systems, which in turn lead to low efficiency in the system parameter correction process and ultimately poor results in patient behavior assessment, this solution introduces an adaptive variable order factor, dynamically adjusts based on an iterative process, enhances search efficiency, introduces a high-order update strategy, improves the accuracy and stability of system parameter optimization, avoids falling into local optimality through perturbation factors, and thereby improves the system's processing speed for psychiatric patient behavior data, ensuring that the accuracy of the system's patient behavior assessment is sufficiently high. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 A schematic diagram of a process flow of a psychiatric patient behavior assessment system provided by the present invention;
[0031] Figure 2 Schematic diagram of the process of the adaptive joint clustering module.
[0032] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION
[0033] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0034] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.
[0035] Example 1, see Figure 1 , a psychiatric patient behavior assessment system provided by the present invention includes a data acquisition module, a behavior data matrix construction module, an adaptive joint clustering module and a psychiatric patient behavior assessment module;
[0036] The data collection module collects historical mental patient behavior data and real-time mental patient behavior data; and sends the data to the behavior data matrix construction module;
[0037] The behavior data matrix construction module receives data sent by the data collection module; constructs a behavior data matrix based on the collected data; and sends the data to the adaptive joint clustering module;
[0038] The adaptive joint clustering module receives the data sent by the behavior data matrix construction module; introduces hidden variables to perform row clustering on the behavior data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and performs adaptive joint clustering. After clustering correction, the parameters are updated by introducing a dynamic learning rate; an adaptive variable order factor is introduced, dynamically adjusted based on an iterative process, and a high-order update strategy is introduced to optimize the initial parameters of the model, thereby realizing clustering of the behavior data; and the data is sent to the psychiatric patient behavior assessment module;
[0039] The psychiatric patient behavior assessment module receives data sent by the adaptive joint clustering module; and implements the assessment of real-time psychiatric patient behavior data based on the clustering results.
[0040] Example 2, see Figure 1 This embodiment is based on the above embodiment. In the data collection module, both the historical mental patient behavior data and the real-time mental patient behavior data include emotional state, sleep duration, social interaction, physiological indicators, physical activity level and anxiety level; the historical mental patient behavior data also includes behavioral risk level; the behavioral risk level is used as a data label, and the label is only used to define the clustering label and does not participate in the clustering process.
[0041] Example 3, see Figure 1This embodiment is based on the above embodiment. The behavior data matrix construction module converts the collected data into vectors and performs standardization processing, and then constructs a matrix. The rows of the matrix represent patients, and the columns of the matrix represent the behavioral characteristics of the patients; the behavior data matrix is obtained; and the clustering label is defined as the label with the largest number of historical data in the cluster.
[0042] Example 4, see Figure 1 and Figure 2 Based on the above embodiment, this embodiment includes the following contents:
[0043] Row clustering unit: Row clustering is to cluster the rows of the data matrix, cluster the patient rows into H behavioral risk level groups based on the Gaussian mixture model, and identify the differences in behavioral risk levels between different patient groups. The probability of row clustering is expressed as: ; The distribution of hidden variables is expressed as: ; The distribution of the behavioral data of mental patients in the context of row clustering is expressed as: ; where p(·) is the probability; p(·|·) is the conditional probability; vhr is the probability that patient r belongs to behavioral risk level h; τ h is the population proportion of the hth cluster, indicating the proportion of patients belonging to this cluster; whr is a hidden variable vector, indicating the hidden variables associated with patient r; N(·) is a Gaussian distribution, is the identity matrix; mr is the behavioral data vector of patient r; is the mean vector of the h-th cluster, representing the feature center of the cluster; is the weight loading matrix, describing the relationship between behavioral features and hidden variables; is the diagonal covariance matrix of the hth cluster, representing the variance of independent errors; H is the total number of behavioral risk level groups;
[0044] Column clustering unit; row clustering is to cluster the columns of the data matrix, cluster the behavioral features into L feature categories based on hierarchical clustering to identify feature groups with similar behavioral risk levels, and define the behavioral feature clustering as ; Calculate the covariance matrix after column clustering, expressed as: ;in, is an indicator matrix, which is 1 if the rth row belongs to the lth column cluster, and 0 otherwise; is the covariance matrix after column clustering; is a diagonal matrix containing the characteristic covariance of column clustering; T is the matrix transpose;
[0045] Adaptive joint clustering unit; combining row clustering and column clustering, the probability density of the patient's behavioral data is finally expressed as: ;
[0046] Cluster correction unit: updates the posterior probability of patient row clustering, expressed as: ; Update the posterior probability of the hidden variable, expressed as: ; The model parameters are optimized by maximizing the logarithmic expectation, expressed as: ; Where E[·] is the expected value of the behavioral risk level; Maximize is the maximization operation; p(·; ·) is the joint distribution probability; are model parameters; is the initial parameter;
[0047] Parameter updating unit; update population ratio, expressed as: ; Update the mean vector, expressed as: ; Update the covariance matrix, expressed as: ;in, and are the population proportions of the h-th cluster after and before the update, respectively; and are the mean vectors after and before the update respectively; and are the covariance matrices before and after the update; α is the dynamic learning rate, , α 0 is the initial learning rate, T1 is the maximum number of iterations, is a smoothing term; R is the total number of patients;
[0048] Clustering determination unit; the maximum number of iterations and clustering threshold are pre-set. When the average silhouette coefficient of all samples is higher than the clustering threshold when clustering converges, clustering is completed; otherwise, if the maximum number of iterations is reached, the initial parameters are corrected; otherwise, the adaptive joint clustering unit is returned to continue iteration;
[0049] Initial parameter correction unit.
[0050] By performing the above operations, the general psychiatric patient behavior assessment system is unable to capture the potential complex relationships in the behavioral data of psychiatric patients, and is only limited to observing surface data, which makes the system poorly adaptable to the complex behavioral data of patients, and thus leads to low accuracy of behavioral assessment. This solution introduces hidden variables to perform row clustering on the behavioral data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and uses the correlation between patient behavioral characteristics and behavioral risk levels based on adaptive joint clustering to improve the system's robustness to behavioral data and the accuracy of behavioral assessment results. The dynamic learning rate update parameter is introduced to improve the convergence of the system when processing psychiatric patient behavior data, thereby enabling the system to evaluate patient behavior more quickly.
[0051] Example 5, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In the adaptive joint clustering module, the initial parameter correction unit specifically includes:
[0052] Construct parameter space; construct parameter search space based on initial hidden variable vector, mean vector, covariance matrix, and weight loading matrix; randomly initialize the search individual position, and use the average silhouette coefficient of all samples when clustering convergence is reached based on the individual position as the individual fitness value;
[0053] Position update; introduce adaptive variable order factor; the formula used is as follows:
[0054] ;
[0055] ;
[0056] ;
[0057] Where v is the adaptive variable order factor at the tth iteration; t is the current iteration number, and tmax is the maximum iteration number; is the position of the jth dimension of the uth individual; f u is the individual fitness value; is the position of the jth dimension of the global optimal individual; γ is the attenuation coefficient; γ0 is the initial attenuation coefficient; is the disturbance factor; randn is a standard normal distribution random number; is the decay rate;
[0058] Search judgment; when there is an individual fitness value higher than the clustering threshold, the initial parameter correction is completed, the individual position is the optimal initial parameter, and the clustering result is obtained; if the maximum number of searches is reached, return to the constructed parameter space: otherwise update the fitness value and return the position update.
[0059] By performing the above operations, the general psychiatric patient behavior assessment system has the problems of improper initial system parameter settings, low search efficiency during parameter adjustment, weak global exploration capabilities, and thus low efficiency of the system parameter correction process, and ultimately poor effect on patient behavior assessment. This solution introduces an adaptive variable order factor, dynamically adjusts based on an iterative process, enhances search efficiency, introduces a high-order update strategy, improves the accuracy and stability of system parameter optimization, avoids falling into local optimality through perturbation factors, and thereby improves the system's processing speed for psychiatric patient behavior data, ensuring that the final system's accuracy in patient behavior assessment is high enough.
[0060] Example 6, see Figure 1This embodiment is based on the above embodiment. The psychiatric patient behavior assessment module is based on the clustering result, and uses the cluster label to which the real-time psychiatric patient behavior data belongs as the assessment result of the real-time psychiatric patient behavior data.
[0061] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0062] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.
[0063] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.
Claims
1. A psychiatric patient behavior assessment system, characterized in that: The system includes a data acquisition module, a behavioral data matrix construction module, an adaptive joint clustering module, and a psychiatric patient behavior assessment module; The data collection module collects historical mental patient behavior data and real-time mental patient behavior data; And send the data to the behavioral data matrix building module; The behavior data matrix construction module receives data sent by the data acquisition module; Build a behavior data matrix based on the collected data; and send the data to the adaptive joint clustering module; The adaptive joint clustering module receives the data sent by the behavior data matrix construction module; introduces hidden variables to perform row clustering on the behavior data based on the Gaussian mixture model, performs column clustering based on hierarchical clustering, and performs adaptive joint clustering. After clustering correction, the parameters are updated by introducing a dynamic learning rate; an adaptive variable order factor is introduced, dynamically adjusted based on an iterative process, and a high-order update strategy is introduced to optimize the initial parameters of the model, thereby realizing clustering of the behavior data; and the data is sent to the psychiatric patient behavior assessment module; The psychiatric patient behavior assessment module receives data sent by the adaptive joint clustering module; Based on the clustering results, the real-time behavioral data of mentally ill patients can be evaluated; The adaptive joint clustering module includes an adaptive joint clustering unit; row clustering and column clustering are combined, and the probability density of the patient's behavior data is finally expressed as: ; where p(·|·) is the conditional probability; vhr is the patient r belonging to the behavioral risk level h; mr is the behavioral data vector of patient r; H is the total number of behavioral risk level groups; τ h is the population proportion of the hth cluster, indicating the proportion of patients belonging to this cluster; N(·) is a Gaussian distribution, is the identity matrix; is the mean vector of the h-th cluster, representing the feature center of the cluster; is the diagonal covariance matrix of the h-th cluster, representing the variance of the independent errors.
2. A psychiatric patient behavior assessment system according to claim 1, characterized in that: The adaptive joint clustering module specifically includes the following contents: Row clustering unit: Row clustering is to cluster the rows of the data matrix, and cluster the patient rows into H behavioral risk level groups based on the Gaussian mixture model. The probability of row clustering is expressed as: ; The distribution of hidden variables is expressed as: ; The distribution of the behavioral data of mental patients in the context of row clustering is expressed as: ; where p(·) is the probability; p(·|·) is the conditional probability; vhr is the probability that patient r belongs to behavioral risk level h; τ h is the population proportion of the hth cluster, indicating the proportion of patients belonging to this cluster; whr is a hidden variable vector, indicating the hidden variables associated with patient r; N(·) is a Gaussian distribution, is the identity matrix; mr is the behavioral data vector of patient r; is the mean vector of the h-th cluster, representing the feature center of the cluster; is the weight loading matrix, describing the relationship between behavioral features and hidden variables; is the diagonal covariance matrix of the hth cluster, representing the variance of independent errors; H is the total number of behavioral risk level groups; Column clustering unit; row clustering is to cluster the columns of the data matrix, cluster the behavioral features into L feature categories based on hierarchical clustering, and define the behavioral feature clustering as ; Calculate the covariance matrix after column clustering, expressed as: ;in, is an indicator matrix, which is 1 if the rth row belongs to the lth column cluster, and 0 otherwise; is the covariance matrix after column clustering; is a diagonal matrix containing the characteristic covariance of column clustering; T is the matrix transpose; Adaptive joint clustering unit; Cluster correction unit: updates the posterior probability of patient row clustering, expressed as: ; Update the posterior probability of the hidden variable, expressed as: ; The model parameters are optimized by maximizing the logarithmic expectation, expressed as: ; Where E[·] is the expected value of the behavioral risk level; Maximize is the maximization operation; p(·; ·) is the joint distribution probability; are model parameters; is the initial parameter; Parameter updating unit; update population ratio, expressed as: ; Update the mean vector, expressed as: ; Update the covariance matrix, expressed as: ;in, and are the population proportions of the h-th cluster after and before the update, respectively; and are the mean vectors after and before the update respectively; and are the covariance matrices before and after the update; α is the dynamic learning rate, , α0 is the initial learning rate, T1 is the maximum number of iterations, is a smoothing term; R is the total number of patients; Clustering determination unit; the maximum number of iterations and clustering threshold are pre-set. When the average silhouette coefficient of all samples is higher than the clustering threshold when clustering converges, clustering is completed; otherwise, if the maximum number of iterations is reached, the initial parameters are corrected; otherwise, the adaptive joint clustering unit is returned to continue iteration; Initial parameter correction unit.
3. A psychiatric patient behavior assessment system according to claim 2, characterized in that: The initial parameter correction unit specifically includes: Construct parameter space; construct parameter search space based on initial hidden variable vector, mean vector, covariance matrix, and weight loading matrix; randomly initialize the search individual position, and use the average silhouette coefficient of all samples when clustering convergence is reached based on the individual position as the individual fitness value; Position update; introduce adaptive variable order factor; the formula used is as follows: ； ； ； Where v is the adaptive variable order factor at the tth iteration; t is the current iteration number, and tmax is the maximum iteration number; is the position of the jth dimension of the uth individual; f u is the individual fitness value; is the position of the jth dimension of the global optimal individual; γ is the attenuation coefficient; γ0 is the initial attenuation coefficient; is the disturbance factor; randn is a standard normal distribution random number; is the decay rate; Search judgment; when there is an individual fitness value higher than the clustering threshold, the initial parameter correction is completed, the individual position is the optimal initial parameter, and the clustering result is obtained; if the maximum number of searches is reached, return to the constructed parameter space: otherwise update the fitness value and return the position update.
4. A psychiatric patient behavior assessment system according to claim 1, characterized in that: In the data collection module, the historical psychiatric patient behavior data and the real-time psychiatric patient behavior data both include emotional state, sleep duration, social interaction, physiological indicators, physical activity level and anxiety level; the historical psychiatric patient behavior data also includes behavioral risk level; the behavioral risk level is used as a data label, and the label is only used to define the clustering label and does not participate in the clustering process.
5. A psychiatric patient behavior assessment system according to claim 1, characterized in that: The psychiatric patient behavior assessment module is based on clustering results, and uses the cluster label to which the real-time psychiatric patient behavior data belongs as the assessment result of the real-time psychiatric patient behavior data.
6. A psychiatric patient behavior assessment system according to claim 1, characterized in that: The behavior data matrix construction module converts the collected data into vectors and performs standardization processing, and then constructs a matrix, in which the rows of the matrix represent patients and the columns of the matrix represent the behavioral characteristics of the patients; obtains the behavior data matrix; and defines the clustering label as the label with the largest number of historical data in the cluster to which it belongs.